Single agents have a reliability ceiling. As context grows and complexity compounds, requirements get buried, assumptions go unchallenged, and shortcuts become harder to catch. The agent can optimize for completion while losing track of what correctness requires.
Independent, adversarial review creates a way through that ceiling. A reviewer starts with fresh context and looks for reasons the work should fail. Rejection turns those failures into concrete repair tasks. Each revision faces renewed scrutiny, allowing the system to recover from mistakes that would otherwise survive all the way to “done.”
zeroshot is a multi-agent framework built around this cycle. Define your graph topology, choose your agents and models, and make independent approval a requirement for delivery. The engine enforces the process.
This graph adds a security review when a change touches authentication, payments, or migrations:
worker ◄─────────── any "no" ────┐
▼ │
risk check │
├─ auth, payments, migration? │
│ ▼ │
│ security review ─────── no ─┤
▼ ▼ │
acceptance + code review ─── no ─┘
▼ both yes
deliver
Build your own graph or adapt a template, then save it as a reusable profile.
See more graph examples on zeroshot.sh.
npm install -g @the-open-engine-company/zeroshotThis installs skills for Claude Code, Codex, Copilot CLI and Pi. Open your agent in a repo and ask:
design a zeroshot graph for this repo
It reads the repo and helps you choose the graph, models, and delivery. Then plan a task:
plan this with zeroshot: let users export their data as CSV
Run the plan:
zeroshot it
Watch with zeroshot ui. Runs happen locally using your agent's existing login, subscriptions
included. See installation options to skip skill installation.
Export a template with Git delivery disabled:
zeroshot template show software-change --delivery none > graph.jsonEdit the graph, choose a runtime, and run it. The review loop guide has complete graph, input, and runtime files; the CLI quick start shows validation and execution.
Run locally, self-host with Docker, or share graphs and run on zeroshot Platform using your own model API keys.
I used Zeroshot for my last two contracts: a production line automation and control system, and a multi-host sync and admin system for Webflow and Guesty. Very happy with it!
— Mark Dingwall
you guys are solving this in a way that is more satisfying than the google ADK (to me) and it’s fuckin google.
— Fav
- Graph contract and execution
- CLI reference, Python SDK, and Cluster API
- Delivery:
software-changedefaults to PR delivery. Runs can also stay local, push a branch, or merge. - Active Memory
Development guide · UI development · Publishing · Security · Issues · Discord
MIT. See LICENSE.
